Benchmark
MLEB - Bar Exam QA
Embedding benchmarks measure how well models find related items, for example when matching a search to relevant results. Scores are percentages.
- Models measured
- 18
- Scores recorded
- 18
- Sources
- 1
- IQ weight
- 0.80
Score distribution
- 50.1%
- Mean
- 53.4%
- Median
- 9.3%
- Std dev
- 38.6%
- 10th percentile
- 60.5%
- 90th percentile
- 34.7%
- Min
This benchmark has a weight of 0.80 for embedding models, measured over 18 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | Score |
|---|---|---|
| 1 | voyage-4-largeBar Exam QA· MLEB data | 66.2% |
| 2 | voyage-4Bar Exam QA· MLEB data | 62.7% |
| 3 | Qwen3 Embedding 8BBar Exam QA· MLEB data | 60.5% |
| 4 | Gemini Embedding 001Bar Exam QA· MLEB data | 57.8% |
| 5 | Qwen3 Embedding 4BBar Exam QA· MLEB data | 56.6% |
| 6 | voyage-4-liteBar Exam QA· MLEB data | 55.9% |
| 7 | Qwen3 Embedding 0.6BBar Exam QA· MLEB data | 55% |
| 8 | Jina Embeddings v5 Text SmallBar Exam QA· MLEB data | 54.4% |
| 9 | Text Embedding 3 LargeBar Exam QA· MLEB data | 53.4% |
| 10 | Jina Embeddings v5 Text NanoBar Exam QA· MLEB data | 53.3% |
| 11 | Jina Embeddings v4Bar Exam QA· MLEB data | 46.9% |
| 12 | snowflake-arctic-embed-l-v2.0Bar Exam QA· MLEB data | 45.8% |
| 13 | snowflake-arctic-embed-m-v2.0Bar Exam QA· MLEB data | 45.3% |
| 14 | Text Embedding 3 SmallBar Exam QA· MLEB data | 39.6% |
| 15 | BGE M3Bar Exam QA· MLEB data | 39.1% |
| 16 | Text Embedding Ada 002Bar Exam QA· MLEB data | 38.6% |
| 17 | Multilingual E5 Large InstructBar Exam QA· MLEB data | 36.8% |
| 18 | Granite Embedding Small English R2Bar Exam QA· MLEB data | 34.7% |
Data credits: MLEB (Isaacus), MIT (licence)